commit | 5b571b74228d2c3fbc139c3532bcb9a33293ba7f | [log] [tgz] |
---|---|---|
author | Ken MacKay <kmackay@gmail.com> | Thu Jun 25 22:41:59 2015 -0700 |
committer | Ken MacKay <kmackay@gmail.com> | Thu Jun 25 22:41:59 2015 -0700 |
tree | 2dfd45daccd828491f4565961a31499ca5d877cb | |
parent | bbad8797486d2fb07788e6ba4422e65215fff5c2 [diff] |
Regularize the bitcount for the private key. This is so that attackers cannot use a side channel attack to learn the number of leading zeros. This causes about 1% slower keypair generation and ECDH. Note: this is not done for secp160r1, since it would have a larger performance impact (about 2% slower on average) and requires the vli_xxx_n functions, leading to a significant increase in code size.
A small and fast ECDH and ECDSA implementation for 8-bit, 32-bit, and 64-bit processors.
The old version of micro-ecc can be found in the “old” branch.
Compressed points are represented in the standard format as defined in http://www.secg.org/collateral/sec1_final.pdf; uncompressed points are represented in standard format, but without the 0x04
prefix. uECC_make_key()
, uECC_shared_secret()
, uECC_sign()
, and uECC_verify()
only handle uncompressed points; you can use uECC_compress()
and uECC_decompress()
to convert between compressed and uncompressed point representations.
Private keys are represented in the standard format.
I recommend just copying (or symlink) uECC.h, uECC.c, and the appropriate asm_<arch>_.inc (if any) into your project. Then just #include "uECC.h"
to use the micro-ecc functions.
For use with Arduino, you can just create a symlink to the uECC
directory in your Arduino libraries
directory. You can then use uECC just like any other Arduino library (uECC should show up in the Sketch=>Import Library submenu).
See uECC.h for documentation for each function.
uECC_CURVE
and uECC_ASM
, you must change them in your Makefile or similar so that uECC.c is compiled with the desired values (ie, compile uECC.c with -DuECC_CURVE=uECC_secp256r1
or whatever).uECC_ASM
is defined to uECC_asm_small
or uECC_asm_fast
), you must use the -fomit-frame-pointer
GCC option (this is enabled by default when compiling with -O1
or higher).uECC_ASM
is defined to uECC_asm_fast
), you must use the -fomit-frame-pointer
GCC option (this is enabled by default when compiling with -O1
or higher).-O1
or higher).advapi32.lib
system library.All tests were built using gcc 4.8.2 with -O3
, and were run on a Raspberry Pi B+. uECC_ASM
was defined to uECC_asm_fast
and ECC_SQUARE_FUNC
was defined to 1
in all cases. All times are in milliseconds.
All tests were built using avr-gcc 4.8.1 with -Os
, and were run on a 16 MHz ATmega256RFR2. Code size refers to the space used by micro-ecc code and data.
In these tests, uECC_ASM
was defined to uECC_asm_fast
and ECC_SQUARE_FUNC
was defined to 1
in all cases.
In these tests, uECC_ASM
was defined to uECC_asm_small
and ECC_SQUARE_FUNC
was defined to 0
in all cases.
In these tests, uECC_ASM
was defined to uECC_asm_fast
and ECC_SQUARE_FUNC
was defined to 1
in all cases.
In these tests, uECC_ASM
was defined to uECC_asm_small
and ECC_SQUARE_FUNC
was defined to 0
in all cases.